Tau PET Links Some Midlife Hallucinations and Delusions to Neurodegeneration

In November 1901, a 51-year-old woman named Auguste Deter entered a psychiatric hospital in Frankfurt. She suspected her husband, heard and saw things others did not, lost words and memories, and became disoriented. Her physician, Alois Alzheimer, resisted forcing her into a simple choice between a “mental” disease and a disease of the brain. After her death in 1906, he examined her tissue and found the abnormalities later known as amyloid plaques and neurofibrillary tangles.

The modern history of dementia thus began with a patient whose story included delusions and hallucinations. Yet medicine spent much of the next century building psychiatry, which treated disturbances of thought and perception, and neurology, which treated degenerating nerve cells, as neighboring but separate houses. Patients’ symptoms never respected the property line.

A Japanese study published August 3, 2026, in Molecular Psychiatry has opened a small window through that wall. Researchers from the National Institutes for Quantum Science and Technology (QST), Kyoto University, Keio University, the Institute of Science Tokyo and collaborating hospitals compared 37 people whose first psychotic illness began after age 40 with 47 age-matched, cognitively unimpaired controls. Every participant underwent imaging for amyloid and tau.

The rate of tau positivity was about 4.4 times higher in patients. Read alone, that number sounds like a single answer. The deeper result was heterogeneity. The patients divided into three nearly equal groups: 13 with an Alzheimer-like amyloid-positive profile, 12 who were amyloid-negative but tau-positive, and 12 negative for both markers.

What the researchers actually did

The participants were recruited from July 2017 through October 2023. Thirteen patients had developed psychosis between ages 40 and 60, while 24 had developed it after 60. The group included schizophrenia, schizoaffective disorder, delusional disorder, acute and transient psychotic disorder, and major depression with psychotic features.

The investigators excluded people with cognitive impairment or known neurological disease at onset, psychiatric illness before 40, severe traumatic brain injury or substance abuse. At least two board-certified psychiatrists and a board-certified neurologist assessed the patients. Controls scored at least 28 on the Mini-Mental State Examination.

Pooling multiple diagnoses under the syndrome of late-onset psychosis is both a strength and a limit. It lets a study ask what different people with a clinically similar late-life presentation might share. With only 37 patients, however, it cannot establish a reliable pathology rate for each diagnosis.

Of the 13 amyloid-positive patients, 12 were also tau-positive. All 13 had developed psychosis after age 60. That strengthens the possibility that some very-late-onset cases reflect a prodromal or preclinical phase of Alzheimer’s disease. Yet half of the 24 amyloid-negative patients were also tau-positive. Their patterns varied markedly, appearing in parietal, occipital and frontal regions, the precentral gyrus and the brainstem.

Tau is more than an Alzheimer’s marker

Tau normally binds to microtubules, part of the internal scaffolding that helps neurons maintain their shape and transport material. When tau becomes abnormally modified and misfolded, it can accumulate inside cells. Alzheimer’s disease is the best-known tau disorder, but it is far from the only one.

Different combinations and conformations of three-repeat and four-repeat tau are associated with progressive supranuclear palsy, corticobasal degeneration, Pick’s disease, argyrophilic grain disease and other tauopathies. Florzolotau—the second-generation tracer used in this study—was designed to detect a broader range of Alzheimer and non-Alzheimer tau deposits than earlier tracers.

That makes the 12 amyloid-negative, tau-positive patients especially intriguing. It also makes them the group that demands the most restraint. A colored PET image is not a pathology slide. It indirectly measures where a radiotracer binds, and the image may include physiological or off-target binding. The scan could not determine which specific tauopathy each person had. Longitudinal data, other biomarkers and eventual imaging-to-autopsy comparisons will be needed.

From microscope to molecular camera

Almost four decades separated tau’s discovery in 1975 from convincing maps of the protein in living human brains. PET uses a tiny amount of a radioactive tracer that binds to a molecular target. Detectors capture the signal produced when emitted positrons annihilate with electrons. Unlike MRI, which primarily reveals structure, PET is a chemical camera: it estimates where a particular biological process is taking place.

In 2013, researchers in the QST lineage reported the tracer PBB3 as a way to visualize tau in living people. That work ultimately led to florzolotau, also known as ¹⁸F-PM-PBB3 or ¹⁸F-APN-1607, developed for clearer imaging across several tau pathologies. The new study turned that improved camera toward one of psychiatry’s oldest questions.

Why onset after 40 matters

Schizophrenia most commonly begins in adolescence or young adulthood. When severe delusions or hallucinations first appear after 40—especially after 60—clinicians broaden the search. Primary psychiatric illness remains possible, but so do adverse drug effects, steroid exposure, sensory impairment, seizures, autoimmune or infectious disease, endocrine and metabolic disorders, tumors, vascular injury and neurodegeneration.

An international consensus in 2000 proposed “late-onset schizophrenia” for onset between 40 and 60, and “very-late-onset schizophrenia-like psychosis” for onset after 60. Late presentations have often been associated with prominent persecutory delusions and “partition” beliefs that people, forces or materials can pass through walls, alongside fewer negative symptoms and less formal thought disorder than typical early-onset schizophrenia.

Epidemiological work has found a higher subsequent risk of dementia among people with very-late-onset psychosis. Autopsy studies have reported argyrophilic grain disease, Lewy body disease, corticobasal degeneration, progressive supranuclear palsy and other pathologies. Autopsy, however, shows only the end of the story. The new PET study matters because it found molecular signals in living participants who were not cognitively impaired when their psychiatric illness began.

What correlated—and what did not

At the group level, patients had a higher tau signal than controls in the parietal cortex. Among amyloid-negative participants, patients had higher signals in parietal and occipital regions. In the amyloid-positive subgroup, more parietal tau was associated with a lower Frontal Assessment Battery score, a measure that probes planning, shifting attention and inhibitory control. The correlation was moderate (r = −0.58) and just below the conventional threshold for statistical significance (p = 0.04).

The negative finding is just as important. The study did not find a significant association between PET signal and the presence or history of psychosis-related clinical symptoms. None of the patients had visual hallucinations, so this was not simply a study of the classic hallucination profile seen in dementia with Lewy bodies. Tau distinguished biological groups, but it did not function as a dial measuring an individual’s delusions or hallucinations.

That need not be surprising. A symptom can emerge from the interaction of pathology, neural circuits, cognition, medicines, sensory loss, stress and social conditions. If so, the amount of one protein will not line up neatly with the intensity of lived experience. The study’s contribution is to expose biological diversity beneath a symptom-defined category.

What this study cannot yet say

Seven of the 47 healthy controls were tau-positive. Age-related pathology and clinically silent changes mean that “positive” cannot simply be translated as “diseased.” PET also carries cost, limited availability, radiation exposure and difficult questions about how to explain uncertain results. Nothing in this study supports sending every person with hallucinations or delusions for a tau scan.

The sample may also reflect referral and selection effects: people who reach a specialized imaging study are not necessarily representative of those treated in general hospitals, living without care in the community or unable to participate. Larger cohorts may revise the percentages substantially.

What would have to happen before care changes

If years of follow-up can show that particular images reliably predict particular cognitive or movement syndromes, late-onset psychosis might eventually be stratified by disease process rather than symptoms alone. An Alzheimer-like amyloid-and-tau profile could lead down one diagnostic and therapeutic pathway; a probable non-Alzheimer tauopathy down another; absence of either marker toward a wider search.

Getting there will require larger multicenter and internationally diverse cohorts; repeated measurement of cognition, movement and everyday function; blood and cerebrospinal-fluid biomarkers; MRI; comparison with alpha-synuclein and other pathological proteins; and eventual correlation with neuropathology. Researchers must also determine whether medications influence tracer measurements or clinical trajectories.

In ordinary care, imaging comes after careful history-taking, physical and neurological examination, medication review, cognitive testing and appropriate laboratory or structural imaging. Sudden confusion, fever, fluctuating consciousness, weakness on one side, seizures, danger to oneself or others, or loss of the ability to meet basic needs calls for urgent assessment—not a research PET scan.

A continuum, not a conquered border

One hundred and twenty-five years after Auguste Deter entered Alzheimer’s clinic, researchers can turn molecules in the living brain into a map of color. That is a profound technical change. It does not mean that an image can read the whole story of a person.

The 37 patients in this study were not miniature versions of one disease. They separated into an Alzheimer-like group, a group with signals suggestive of other tau pathology, and a group unexplained by either PET marker. Psychiatry and neurology have not merged. Their boundary has become more transparent.

The decisive evidence will now have to come from time. Who remains stable? Who develops cognitive or movement changes? Which molecular map, if any, predicts that future accurately enough to help rather than frighten? Until those questions are answered, this study is not the end of diagnosis. It is the beginning of a more precise—and more humble—set of questions.